0
Research Papers: Flows in Complex Systems

# Numerical Simulations and Analysis of a Low Consumption Hybrid Air ExtractorOPEN ACCESS

[+] Author and Article Information
Marc Sanchez, Adrien Toutant, Françoise Bataille

PROMES CNRS UPR 8521,
University of Perpignan Via Domitia,
Tecnosud-Rambla de la Thermodynamique,
Perpignan 66100, France

Contributed by the Fluids Engineering Division of ASME for publication in the JOURNAL OF FLUIDS ENGINEERING. Manuscript received October 10, 2016; final manuscript received July 25, 2017; published online September 11, 2017. Assoc. Editor: Wayne Strasser.

J. Fluids Eng 139(12), 121106 (Sep 11, 2017) (10 pages) Paper No: FE-16-1666; doi: 10.1115/1.4037507 History: Received October 10, 2016; Revised July 25, 2017

## Abstract

Hybrid low pressure air extractors are an economic way to enhance indoor air quality. The evaluation of their energetic performances needs the analysis of flow parameters that is typically done with wind tunnel data and numerical simulations. The purpose of this study is to analyze, numerically and experimentally, the flow and the energetic performances of a hybrid rooftop extractor. This innovative extractor has two main features: it works at low difference of pressure, below 50 Pa, and its fan is placed far above the duct outlet, out of the fluid flow. The hybrid extractor works following three modes of operation: stack effect, Venturi effect, and fan rotation. The two first modes of operation allow large energy saving. To analyze the three modes of operation, three sets of corresponding Reynolds-averaged Navier–Stokes (RANS) simulations are developed. The first one allows us to estimate the pressure drop due to the geometry of the air extractor. The second one is used to check the ability of the extractor to generate a suction into the duct in the presence of wind. The final one involves multiple reference frame (MRF) modeling in order to study the flow when the electric motor drives the fan. The numerical simulation configurations are validated with experimental data. A good behavior of the extractor is found for simulations of stack effect mode and Venturi effect mode. The stack effect and the Venturi effect allows the hybrid extractor to work most of the time without electric power. Finally, energetic comparisons are given.

<>

## Introduction

Wind driven ventilation techniques are a low cost and environmental friendly way to remove foul air from buildings and replace it with outdoor air. There are several air extractor devices for the purpose of removing foul air. These devices vary both in geometry and design configuration [1,2]. Numerical and experimental flow analyses are necessary to understand the behavior of the numerous existing technologies. In literature, several experimental configurations are given to test ventilation devices (see Refs. [3,4], or [5] for rooftop turbines or Ref. [6] for static cowls). Computational fluid dynamics (CFD) is increasingly used to analyze or predict the behavior of wind-driven devices. For example, Hughes and Abdul Ghani [79] studied wind catchers using steady-state Reynolds-averaged Navier–Stokes (RANS) simulations. Serag-Eldin [10] applied steady RANS to investigate the performances of a wind-driven device shaped to increase Venturi effect. RANS simulations can also be used as a component of a multi-objective optimization process [11]. Pfeiffer et al. [12] used experimental data and CFD to model cowl performances. Van Hooff et al. [13,14] and Blocken et al. [15] took advantage of the complementarity of both experimental and computational approaches to investigate the performance of a Venturi-shaped roof for natural ventilation. In Ref. [13], the flow conditions are analyzed focusing on the generated underpressure. RANS simulations using renormalization group k–ϵ model match very well with experimental measurements. In Ref. [15], it is shown that performance of the roof is a consequence of the balance between Venturi-effect and wind-blocking effect. In Ref. [14], the influence of the building width on the performance of the Venturi-shaped roof is investigated. Montazeri et al. [16,17] introduce the study of wind catchers using an open-circuit wind tunnel, smoke visualization and CFD modeling (RANS) to investigate the flow pattern and the dimensionless pressure distribution around and inside the wind-catcher. Good adequation was found between CFD results and wind tunnel measurements. Lien and Ahmed [18] used steady-state multiple reference frame (MRF) RANS simulations to study rooftop turbine ventilator flow. The results follow the same behavior than experimental wind tunnel data. For the same type of studies, Farahani et al. [19] used unsteady RANS with moving meshes and two different turbulence models Reynolds stress model and Realizable k–ϵ. The results follow very well experimental data, especially for the simulations using Reynolds stress model.

The present study concerns a hybrid air extractor running with Venturi effect, stack effect, and electricity. By contrast with the previous studies, the fan of the extractor is placed above the duct and does not obstruct it. Consequently, the Venturi effect, induced by the wind and the stack effect, can be used as principal sources of ventilation. When there is no wind, to keep good air conditions inside the building, an electric fan takes over. Moreover, because the duct is not obstructed, in case of fire with power failure, the smoke is still extracted improving the safety of the residents. In literature, there is no paper gathering stack effect, Venturi effect, and electric modes of operation for a hybrid extractor. Moreover, there is no study concerning a hybrid air extractor with the fan placed out of the flow. In the present paper, we propose to make a study of the flow and a performance analysis of this kind of extractor. On one hand, it is expected that the fan placed out of the flow increases efficiency of stack and Venturi effects. On the other hand, it is expected that it decreases the electric mode efficiency. An important point of this work consists in quantifying the global energetic performances of the present hybrid extractor with typical European meteorological conditions.

Three simulation configuration sets are used to characterize and analyze the hybrid air extractor in the three modes of operation. These numerical simulations will allow to study

• The pressure drop (limiting the flow induced by the stack effect),

• the static suction (induced by the wind), and

• the dynamic suction (induced by the rotation of the fan).

The simulations are compared to experimental data obtained according to EN-13141-5 [20] and ISO-5801 [21] standards. We use these experimental results for a validation purpose.

First, in Sec. 2, the used geometry and the three study configurations are described. In Sec. 3, the experimental setup is described. In Sec. 4, the simulation parameters, such as numerical settings, mesh and boundary conditions are explained for each study configuration. In Sec. 5, the results of the simulations are compared with experimental data. The flow generated by the hybrid extractor is presented for each configuration. In Sec. 6, the results obtained in this study will be used to evaluate the energetic performances of the studied hybrid extractor operating in the meteorological conditions of a typical European city. Section 7 concludes the paper.

## Geometry and Configurations

###### Geometry Description of the Extractor.

The present hybrid extractor has two unusual features. First, it operates at low pressure and consequently involves low energy consumption compared to high pressure systems. Second, its fan is placed far above the duct outlet. Compared to classical heating, ventilation, and air conditioning installations that involve complex ductwork with high pressure drop like 700 Pa, the present paper concerns a low pressure hybrid extractor, i.e., the maximum operating pressure is 50 Pa and the maximum flow rate is 1000 m3 h−1.

The air extractor is based on a static cowl for 250 mm diameter ducts. It is composed of a Venturi cone and of a motor driven fan. The extraction operates with the stack effect, the Venturi effect, but also with the rotation of the fan when the stack effect and the external wind are not effective. As shown in Fig. 1, the typical flat cap of the static cowl has been replaced by a fan protected by a guard and a shell containing the motor is placed at the top. The different parts are held together by thin mounting brackets. The contribution of these mounting brackets is neglected (not shown).

The hybrid air extractor, studied in this paper, works following three modes of operation. The cowl effect mode works with stack effect, the Venturi effect mode works with the wind, and the electric mode works with the rotation of the fan. The two first modes called “natural modes” of operation allow large energy saving when the meteorological conditions are adequate (inside temperature larger than outside temperature, wind). The extractor has a very unusual characteristic: its fan is placed outside of the duct. This configuration should allow good performances in the two first mode of operation (stack effect and Venturi effect) but decreases the fan efficiency.

###### Configurations Tested.

The numerical simulations are based on three experimental configurations used to characterize the performances of the rooftop air extractor. These three setups are described in the following paragraphs. Figure 2 shows the different configurations. U is the velocity of the external air flow in m s−1, Ub the bulk velocity in the duct in m s−1, and Ω is the rotation speed of the fan in rad s−1. For the three setups, the results are expressed as dimensionless quantities. These dimensionless quantities are nondimensionalized by energy per volume unit (pressure) and are plotted as functions of the Reynolds number instead of flow rate to allow comparisons with other studies. The Reynolds number is

Display Formula

(1)$Re=UbDν$

with D the duct diameter, and ν the kinematic viscosity in m2 s−1.

###### Pressure Drop Mode Characterization.

This setup is used to measure the pressure drop induced by an air extractor when it operates with stack effect only (chimney effect). It allows to quantify how much the extractor blocks the air outflow.

In this configuration, the air extractor is put on a cylindrical duct with a length of 6 times its internal diameter. Several bulk velocities, Ub, are set at the inlet of this duct to simulate the stack effect but the flow stays isothermal. The range of these velocity values corresponds to Reynolds numbers between 100,000 and 450,000. The external flow velocity U and the rotation speed Ω are equal to zero. The static pressure is measured inside the duct, three diameters below the air extractor. The difference of static pressure between inside the duct and outside gives the pressure drop induced by the air extractor for a given flow rate. The dimensionless pressure drop ζ can be used to characterize the air extractor. It is written as follows: Display Formula

(2)$ζ=ΔPs0.5ρUb2$

ΔPs = ps − p, with ΔPs, the difference of static pressure taken between the inside of the duct (three diameters below the extractor) and the external pressure, and ps and p the static pressure in the duct three diameters below the extractor and the external pressure.

###### Static Mode Characterization.

This setup is used to quantify the capacity of the air extractor to generate a low-pressure inside the duct. This low-pressure is a consequence of the Venturi effect. As shown in Fig. 2, the static mode setup uses a configuration similar to the pressure drop measurement setup. The main difference is the air flow set around the extractor: U = 8 m s−1. The range of the bulk velocities Ub set at the inlet of the duct corresponds to Reynolds numbers from 0 to 130,000. The fan does not rotate and Ω = 0.

The dimensionless quantity Cext is used to quantify the static extraction of the equipment Display Formula

(3)$Cext=ΔPs0.5ρU∞2$

with ρ the density and U the velocity of the flow in the external domain.

For Cext < 0, the Venturi effect generates a low-pressure inside the duct and is able to compensate a pressure drop of the ductwork depending on the Reynolds number. When Cext > 0, the extractor is resistive and cannot compensate the pressure drop of the ductwork. When Cext = 0, the extractor generates a flow, but there is no pressure drop to compensate.

###### Dynamic Mode Characterization.

Hybrid air extractors, like electric air extractors or fans, have to be tested dynamically to characterize and analyze the effect of the rotation of the fan. Figure 2 shows the configuration for the dynamic mode. The difference with the pressure drop mode is the rotation of the fan of the hybrid extractor induced by the electric motor (Ω ≠ 0). There is no external air flow in the domain, U = 0. Different bulk velocities, Ub, are set at the inlet of the extractor (corresponding to Reynolds numbers from 0 to 45,000). The motor is controlled in tension. In function of the flow rate set in the duct, the rotation speed of the fan can vary (see Table 4). As an usual fan, for each rotation speed, the fan behaves following a flow rate/pressure curve. By imposing the flow rate and the rotation speed, it is possible to measure the difference of pressure induced by the system. Like in the other characterization setup, this difference of pressure is representative of the pressure drop that the air extractor can compensate to maintain the flow rate. The difference of static pressure is taken between inside the duct (three diameters below the extractor) and the external domain. It is expressed as a dimensionless quantity Display Formula

(4)$Kp=−ΔPs0.5ρ(ΩR)2$

with R the fan radius (0.175 m), and Ω the rotation speed of the fan in rad s−1.

## Experimental Setup

Experimental data were collected by an independent agency specialized in air extractor and aerodynamic characterization and certification. The results were produced following EN-13141-5 [20] and ISO-5801 [21] standards. The experimental setup is composed of a test-bench described in Sec. 3.1 and a wind tunnel described in Sec. 3.2.

###### Test Bench.

The test bench is composed of (from upstream to downstream) the following:

• A fan unit with variable flow rate,

• a settling chamber,

• a 250 mm diameter duct with a Prandtl probe for flow rate measurements,

• a 250 mm diameter flexible duct, and

• a 250 mm diameter and 1.5 m length duct with the hybrid extractor at its top.

The flow rate is measured using an approximation of a turbulent velocity profile. The velocity is measured with a Prandtl probe placed in the center of the duct. The maximum uncertainty is ±2%. The pressure is measured by four static pressure probes placed in cross in the duct three diameters below the extractor. The manometers have a maximum of ±1% of measurement uncertainty. The rotation speed of the fan is measured by a stroboscope with ±0.5% of measurement uncertainty.

###### Wind Tunnel.

The wind tunnel used for static mode investigations has a nonguided test section of 2 m of diameter. This configuration allows us to study large size objects without a high obstruction of the wind tunnel. The bulk velocity of the fluid can be chosen between 0.5 m s−1 and 30 m s−1. The turbulent intensity inside the wind tunnel is between 1% and 2%.

###### Uncertainty on the Studied Parameters.

The measurement uncertainty on pressure velocity and rotation speed is never larger than ±2%. As we do not consider directly the measured quantities but the quantities Re, ζ, Cext, and Kp, we will evaluate their corresponding relative errors. By neglecting the measurement error made on the density ρ, we can calculate the following relative errors: Display Formula

(5)$|ΔReRe|=|ΔUbUb|≤2%$
Display Formula
(6)$|Δζζ|=|ΔPsPs|+|2ΔUbUb|≤5%$
Display Formula
(7)$|ΔCextCext|=|ΔPsPs|+|2ΔU∞U∞|≤5%$
Display Formula
(8)$|ΔKpKp|=|ΔPsPs|+|2ΔΩΩ|≤2%$

Generally speaking, the relative errors on the calculated experimental quantities are always inferior to 5%. Experimental data are be used for numerical simulation validation.

## Simulation Parameters

###### Numerical Configuration and Turbulence Modeling.

In this study, the CFD opensource tool used is OpenFOAM 2.2.2. The calculation method uses the semi-implicit method for pressure linked equations [22] steady-state incompressible Navier–Stokes equations solver with a geometric-algebraic multigrid pressure solver.

The turbulence model is the realizable Reynolds stress algebraic equation model developed by Shih et al. [23] (NonLinearKEShih). It is a quadratic nonlinear model based on the k–ϵ two-equations formulation. This model was not optimized for swirling flow [24]. Near the wall, if the first grid point is in the viscous sublayer, a no slip boundary condition is applied and the flow is calculated until the wall. If not, the flow behavior is modeled by classical wall functions [25]. The fluid is incompressible air. The kinematic viscosity is $ν=1.56×10−5 m2 s−1$, and the density is $ρ=1.2 kg m−3$.

For all the simulations presented here, we use central-based second order schemes, on all quantities, including diffusion terms. The residuals on the velocity have to be below 10−2 and below 10−3 for the pressure. Two convergence criteria are analyzed before stopping computation: the residuals of pressure and velocities and the stability of the computed pressure in the duct. As the pressure in the duct is, for us, the most important parameter, the calculation, is stopped after the stabilization of its moving average. However, the stabilization of the averaged pressure does not necessarily imply stabilization of all flowfield features.

For pressure drop configuration and static configuration simulations, under-relaxation factors are 0.3 for pressure and 0.7 for velocity and turbulence variables. The convergence of the calculation and the stabilization of the pressure is reached after approximately 3000 iterations.

With under-relaxation factors of 0.3 for pressure and 0.7 for velocity and turbulence variables, dynamic mode simulations have more difficulties to reach convergence. The pressure values can oscillate of ±5 Pa. Central-based differencing schemes are prone to stability issues and might be a contributor of these pressure oscillations. Consequently, the under-relaxation factors were set to 0.2 for pressure and 0.5 for velocity and turbulence variables. The pressure oscillates of ±2 Pa. A moving average is calculated on a sliding interval of 200 iterations. This stable average is used to characterize the extractor. The stabilization of the averaged pressure is reached after 8000–12,000 iterations for dynamic mode simulations.

###### Mesh and Boundary Conditions.

Boundary layer mesh is created on the wall of the geometry. Because the mounting brackets are very thin, we assume their contribution to the flow is very small. This assumption decreases considerably the number of cells necessary to the mesh.

Two mesh studies are presented. The first one concerns the pressure drop mode. The pressure drop mode mesh and the static mode mesh are quite similar; consequently, no mesh study is presented for static mode. The second mesh study concerns the dynamic mode. The phenomena are more complex and require a finer mesh inside the MRF volume.

For the pressure drop mode, three meshes are analyzed, a coarse one composed of 2 million cells, an intermediate one composed of 3.6 cells and a fine one composed of 7.2 million cells. In Table 1, the number of cells, the y* values (maximum, minimum, and average) and the relative difference between experimental data and simulations for static pressure are provided in percent for the static mode at the bulk velocity in the duct Ub of 4 m/s (Re = 62,820). y* is the wall unit defined as

Display Formula

(9)$y*=ρCμ1/4kw1/2ywμ$

with ρ the density, yw the height of the first cell against the wall, kw the turbulent kinetic energy at the wall, and Cμ = 0.09. With the coarse mesh, the relative error is 31%, and it is 23% with the intermediate mesh and finally 12% with the fine mesh. The finer mesh will be used for the computations as it is a good compromise between computational cost and accuracy.

For the dynamic mode, the grid dependence was analyzed, especially in the MRF volume. As presented in Table 2, three meshes were tested, a coarse one, an intermediate one, and a fine one, with 8, 9.4, and 14 million cells, respectively. The relative error on the nondimensional pressure decreases with the number of cells. Grid independence is not reached, but the 14 million cells mesh is already quite computationally intensive and represents yet a good compromise between time calculations and accuracy on global quantities. We keep in mind that a reasonable match in global quantities does not necessarily imply correctness in local quantities.

For this reason, as illustrated in Fig. 3, we analyzed the vertical velocity inside the MRF volume for three different grid densities. The figure shows cut planes across the fan for different grids. Each cross represents a cell. The coarse mesh is on the left, the intermediate mesh is at the top, and the finer mesh is on the right. When the density of cells is high, the crosses are not distinguishable. Globally, the vertical velocity around the blades does not look very grid sensitive. However, the lack of experimental reference data implies that we cannot confirm the correctness of the vertical velocity.

Table 3 provides for every grid the y* values (minimum, maximum, and average) at Ub maximum. For all the simulation configurations, the mesh is quite fine. The y* averaged values are lower or equal to seven for the parts of the extractor. For the duct, they are lower or equal to 22.

The domain size and the boundary conditions are given in Fig. 2. The cubic domain designated by the bold bounds is for the pressure drop mode and dynamic mode. The parallelelipedic domain designated by the thin bounds is for the static mode. The black thin bounds are for the static mode. The fixed parameters are outlined by the arrows and an adjacent symbol. The numbers inside brackets correspond to the different configurations: number 1 for pressure drop, number 2 for static, and number 3 for dynamic.

For each case, the pressure at the external boundaries of the domain is set to zero. At the fluid/solid interface, a classical wall law is set when y* > 11 and a nonslip condition is applied in the other cases.

###### Pressure Drop Mode Configuration.

The mesh is composed of 5.1 million cells. The large number of cells is due to the size of the extractor geometry and to the sharpness of the fan blades and of the guard. The averaged cell size in wall unit of this mesh is $yavg*=5$. Cut sections of the mesh are shown in Fig. 4.

The domain is a cube of 4 m edges. Velocity is set at the inlet of the duct.

###### Static Mode Configuration.

For the static mode, the mesh is composed of 7 million cells. The averaged cell size in wall unit of this mesh is $yavg*=6$. The cells near the blades and the guard are the smallest ones.

Because of the external flow, the domain needs to be larger than for the pressure drop mode. The length of the domain is equal to 8 m, the width is equal to 4 m, and the height to 4 m. A mean profile is set for U, k, and ϵ at the inlet of the duct. At the inlet of the domain the velocity is set to 8 m s−1. The extractor is placed at 3 m from the external flow inlet.

###### Dynamic Mode Configuration.

For the dynamic mode, like for the pressure drop mode, the calculation domain is a cube of 4 m edges. The mesh is composed of 14 million cells. The averaged cell size in wall unit of this mesh is $yavg*=2.6$. The complexity of the phenomena taking place into the extractor needs a very fine mesh in the area between the fan and the cone.

To simulate the rotation of the blades, we used a multiple reference frame modeling. The MRF method has successfully been used with OpenFOAM by Liu et al. [26] on a double blades pump. The rotor is fixed, and a cylindrical rotating reference volume is defined around the blades of the fan. In this volume, the steady-state Navier–Stokes equations become

• incompressibility hypothesis Display Formula

(10)$∇·ur=0$

• momentum equation Display Formula

(11)$∇·(ur⊗ur)+2Ω∧ur+Ω∧Ω∧r=−∇(pρ)+ν∇·∇ur$
with $ur$ the relative velocity, $Ω$ the rotation speed, $r$ the radial vector, ⊗ tensor product, $∧$ cross product, ∇ gradient, and ∇⋅ divergence. The second term on the left-hand side of the second equation is the Coriolis acceleration, and the third term, the centrifugal acceleration [27]. At the interface between inertial and rotating frame, the velocity is transformed from rotating to inertial frame by Display Formula
(12)$uI=Ω∧ur$
with $uI$ the inertial frame velocity.

As this method uses a fixed rotor approximation, it is meaningful only for steady flow simulations. It means also that it will not be helpful for noise prediction for which it has been shown that is more suitable to use an unsteady viscous flow approach [2830]. To work properly, the interfaces between static and rotating frame must be surfaces of revolution about the axis of rotation defined for the rotating frame. As the MRF method is for steady flow, it does not take into account the interactions between the blades and a possible complex stator or another rotor. In our case, the static parts are solids of revolution about the axis of rotation defined for the rotating frame and their interaction with the rotor will not depend on their angular setting. Even if the interaction between blades are known to be properly modeled by the MRF approach, the dynamic mesh method gives even more accurate results but increases computational costs [31]. It was shown that at a very low rotation speed, the solution given by the MRF method deviates from the one given by sliding mesh method [32]. Another parameter is the distance between the inertial parts and the rotating parts and their distances with the inertial/rotating interface [33]. As the fan and the guard are very close, this could impact the results.

A mean profile is set for U, k, and ϵ at the inlet of the duct as boundary conditions. A rotation speed is set for the MRF volume. These conditions are summed up in Table 4.

## Results and Analysis

The knowledge of the static pressure inside the duct allows us to check if the air extractor is able to compensate the pressure drop. Compared to Lien and Ahmed [18], who studied only the flow around the extractor, the collected information concerns the velocity and the pressure measured three diameters below the extractor, inside the duct. On the graphs, experimental data are in red stars and simulations in blue squares.

###### Pressure Drop Mode.

The pressure drop mode simulations converged after about 3000 iterations. The results of these simulations are presented in Fig. 5. In this figure, the experimental data of static pressure were reduced by 1.5%. This modification allows one to take into account the potential measurement errors. The graph shows the dimensionless pressure drop against the air flow rate in the duct. The wind tunnel experiments are in stars, and the results of the simulations are in squares. It can be seen that the ζ evolution is well predicted by the simulation. The value of this quantity is lower than 0.05. It corresponds to a good behavior of the extractor, in the case of a pure convective flow. This value can be explained by the configuration of the extractor, especially by the position of the fan which is far from the duct opening. This value confirms that the extractor allows large energy savings in stack effect mode.

Pictures of the flow can be seen in Fig. 6. They correspond to normal cut planes for y = 0 at Re = 188,462. The nondimensional velocity field magnitude U/Ub and the nondimensional pressure field

Display Formula

(13)$Ps/(0.5ρUb2)$

are represented on this figure.

In Figs. 6(a) and 6(b), cut sections of the extractor flow are presented. The most part of pressure drop is contained around the fan. The air flow is blocked by the motor shell and the fan guard. The flow slows down at the outlet of the duct. A part of the fluid is ejected at the top of the extractor, between the shell and the guard. The other part goes down and exits by the area between the Venturi cone and the guard. The low ζ corresponds to the fan placed out of the duct. This choice of conception allows the air to pass through the extractor in stack effect mode without the help of an external power supply.

###### Static Mode.

Figure 7 shows the dimensionless static extraction quantity of the extractor. 3000 iterations are necessary to reach convergence. The graph shows a good agreement between numerical simulations and wind tunnel experiments even if small differences are visible at high Reynolds numbers. The results show also that below Re ≈ 40,000, the external wind allows the extractor to generate suction into the duct that can decrease electric energy consumption. The Cext values are interesting when they are negative, so when the Reynolds number is inferior to 50,000. This corresponds to the flow rates for which the extractor is able, by using the external airflow, to generate a low pressure inside the duct, and consequently, the flow rates for which the static mode is effective.

In Fig. 8, nondimensional velocity field magnitude U/U and nondimensional pressure field

Display Formula

(14)$ΔPs/(0.5ρU∞2)$

are represented. We see that a part of the suction inside the duct is induced by the extractor that behaves as an obstacle to the external flow. This external flow can be considered as a flow past a circular cylinder for a high Reynolds number. Here, the Reynolds number is Display Formula

(15)$Re=U∞D/ν=128,000$

with U the velocity of the external flow in m s−1, D the duct diameter, and ν the kinematic viscosity in m2 s−1.

In Figs. 8(a) and 8(b), the flow inside the extractor is shown. A part of the external airflow, on the left, passes in the extractor between the Venturi cone and the guard. It accelerates just above the duct. Another part of the external airflow, on the top left, is accelerated between the shell and the guard. The area inside the guard is separated in two parts: low pressure on the left and high pressure on the right. A small part of the flow is ejected downstream, on the right, between the shell and the guard, but the largest part is ejected between the cone and the guard.

###### Dynamic Mode.

The dynamic mode simulation is more complex than the two previous modes. The simulations need 8000 iterations to reach convergence. Several rotation speeds from 718 to 829 rpm are set for each flow rate. They are summed up in Table 4.

Figure 9 shows the dimensionless pressure Kp as a function of Reynolds number Re for experiments (stars) and for simulations (squares). The values from simulations are scaled using a fan affinity law [34]

Display Formula

(16)$ΔPtot∝Ω2R2$

to make the value of Kp at Re = 0 accords to the experimental data. According to this affinity law, the flow rate is proportional to the rotation speed and the pressure is proportional to the square of the rotation speed. The scaled speeds are equal to 89% of the original values. The difference on the rotation may be due to the MRF method. In the studied extractor, the blade tips are very close to the guard and the distance between the static and the moving geometries and their distances with the inertial/rotating interface are very short and could impact the results. Using the fan affinity laws, Fig. 9 shows the simulation results are in very good agreement with the experimental data. The behavior and the amplitude of the curves match perfectly.

Figure 10 shows the efficiency as a function of Reynolds number Re for experiments (stars) and for simulations (squares). The extractor efficiency is defined as

Display Formula

(17)$η=QΔPtotWshaft$

Q is the flow rate in $(m3 s−1) and ΔPtot$ is the increase of total pressure Display Formula

(18)$ΔPtot=ΔPs−12ρU2$

with ΔPs the increase of static pressure, ρ the air density, and U the bulk velocity in the duct. Wshaft is the power W.

In the simulations, Wshaft is calculated as follows: Display Formula

(19)$Wshaft=(MP+Mν)Ω$

with MP in (N·m), pressure forces momentum Mν in (N·m) viscous momentum on the fan blades, and Ω in (rad s−1), the fan rotation speed.

For experimental results, they are calculated as follows: Display Formula

(20)$Wshaft=Wmηm$

with Wm the motor electric power consumption and ηm the motor efficiency. The motor efficiency given by manufacturer is about 40%. In Fig. 10, we took into account a rotation speed of 89% of the original one. An important gap is visible between experimental results and simulations. The numerical simulations do not predict fan efficiency well. The original shape of the fan efficiency profile is well reproduced. Particularly, the maximum fan efficiency is obtained for the correct Reynolds number. However, the amplitude of the calculation errors on pressure forces and viscous momentum are added to the errors on pressure and increase the difference with experiments.

Figure 11 shows the relative velocity around the fan. Viewed from below, the rotation of the blades is anticlockwise. The air flow comes from the center of the fan. The velocity is increased along the blades and deflected down following the blade pitch. At the outer radius, a part of the relative velocity goes in the opposite direction to the rotation, with a decreasing velocity. This effect is due to the fan guard that changes the direction of the flow and slows down the air extracted by the fan. It can potentially decrease the extractor performances.

The Fig. 12(a) represents the streamlines below the fan. The streamlines show that the fan generates a swirling flow around. This particularity is due to the position of the fan outside the duct and could generate power losses in electric mode. The swirling flow can be decomposed in three parts as a function of the distance to the center of the duct, r:

• Part A corresponds to r < 4 cm.

• Part B corresponds to 4 cm < r < 14 cm.

• Part C corresponds to 14 cm < r.

In part A, the flow goes upwards and does not rotate. In the intermediate area, part B, the flow is rotating and goes upwards. In the C area, the flow rotates and goes downwards. At the boundary between the two last areas, the flow rotates but stays in the same plane. These three areas can be distinguished more clearly in Fig. 12(b) that shows velocity in a plane between the Venturi cone and the fan (z = 1.67 m). In this figure, the velocity is represented by three-dimensional vectors.

Finally, Fig. 13(a) shows vertical velocity Uz, tangential velocity Uθ, and radial velocity Ur as functions of the radius. The cut is taken at z = 1.67 m, between the cone and the guard for the three velocities mentioned earlier. A fourth velocity, Uz−duct, in dashes, is taken in the duct three diameters below the extractor. A schematic representation of the extractor is given in Fig. 13(b). The letters (a, b, c, d, e, and f) corresponds to the positions of the parts of the extractor depending on the radius. The values of the radius for each letter are summed up in Table 5. Figure 13(a) confirms that, in the area A, the flow goes upwards without rotation. In area B, vertical velocity decreases while the flow rotates. In area C, the vertical velocity becomes negative and reaches the minimal value for r = 0.17 m such as tangential velocity. With r increasing, both velocities increase to reach values close to zero. Radial velocity is zero in the area A, increases slightly in area B, and goes to a negative value in the area C.

Figure 13 gives information about the interactions between the fluid and the geometry. In position (a), the blade height makes the tangential and the vertical velocity decrease. Between r = 0 and r = b, the blades are fixed on a metal disk blocking the air. In this area, the decrease in tangential velocity and the increase in radial velocity is the largest. In r = c, because the disk does not longer interfere with the fluid, the vertical velocity decreases more slightly. In position (d), the changes in the blade shape make the vertical velocity negative, the tangential velocity increases slightly because of the friction with the top of the cone. In position (e), the tip of the blade makes the tangential velocity the lowest, close to the guard, the vertical velocity is minimal. After the end of the guard, in f, the values return to a flat behavior.

## Energetic Evaluation of the Hybrid Air Extractor

In Sec. 5 of this paper, we used numerical simulations to determine the dimensionless pressure drop ζ, the dimensionless static extraction quantity Cext, and the dimensionless dynamic extraction quantity KP. The information gathered in Sec. 5 are now used to evaluate the energetic performances of the present hybrid extractor with the meteorological conditions of a typical European city.

We consider a building needing a minimum flow rate of 300 m3 h−1 with a low pressure ductwork (shunt duct) characterized by a pressure drop of 20 Pa at this flow rate. The mean stack height of the ductwork is H = 8 m. The hybrid air extractor works in natural mode (without motor) when the sum of the negative pressure induced by the stack effect and the Venturi effect is equal or above the needed 20 Pa. Following the experimental data in Fig. 10, at 300 m3 h−1 (or Re = 28,000), the fan efficiency is about 10%. If we take into account the motor efficiency of ηm = 40%, the global efficiency of the extractor is 4%. The working point of ΔPtot = 20 Pa and Q = 300 m3 h−1 corresponds to a fan power of Display Formula

(21)$Wfan=QΔPtot=1.67 W$

The estimated power consumption of the extractor is consequently Display Formula

(22)$Wextractor=Wfanηfanηm=41.67 W$

The flow rate of 300 m3 h−1 corresponds to a Reynolds number of 28,000. As shown in Fig. 5, at this value, the pressure drop is below 0.07, and consequently, the pressure drop ΔPdrop induced by the extractor is Display Formula

(23)$ΔPdrop=ζ12ρUb2<0.23 Pa$

and so negligible. It does not perturb the stack effect. The negative pressure induced by the stack effect is known as Display Formula

(24)$ΔPdrop=ΔρgH$

with the gravity of earth g = 9.8 m s−2. Δρ is the difference of the air density between outside and inside the building. If we consider the air as an incompressible ideal gas, can be expressed as a difference of temperature Δρ, Display Formula

(25)$Δρ=(PatmMairRTout−PatmMairRTin)gH$
Display Formula
(26)$Δρ=(PatmMairRTin−ToutTinTout)gH$

If we consider the temperature of air so much around 293 K (the constant inside temperature), we can simplify the ΔPdrop relation as Display Formula

(27)$ΔPdrop≈0.044(Tin−Tout)H$

As shown in Fig. 7, the dimensionless static extraction quantity at this Reynolds number is Cext = 1.11. The static pressure induced by Venturi effect is Display Formula

(28)$ΔPventuri=Cext12ρUout2$

In pure electric mode, the extractor consumes 365 kWh over a whole year. Following the meteorological conditions for a typical European city, the resulting pressure Display Formula

(29)$ΔPdrop+ΔPventuri$

will be above 20 Pa during 79% of the year. The consumption of the hybrid extractor will be only 76.7 kWh and represents 288.4 kWh of energy saving.

If we replace this extractor by a classical backward centrifugal fan, with ηc = 70% of efficiency [35] and a motor of ηm = 40% of efficiency, like previously, the electric power consumption at the working point of ΔPtot = 20 Pa and Q = 300 m3 h−1 will be Display Formula

(30)$Wfanηcηm=ηcηmQΔPtot=1.67×0.7×0.4=5.96 W$

Over a whole year in pure electric running, the centrifugal fan power consumption will be only 52.2 kWh so, 1.47 times less than the hybrid extractor.

Thereby, the position of the fan (out of the duct) seems to be a good solution to make energy saving. Indeed, the natural mode of the hybrid extractor is used almost 80% of the time. However, the shapes of the fan and the guard have to be improved in order to increase the efficiency of the electric mode to make this extractor competitive compared to a centrifugal fan. Note that improving the guard shape is interesting because, in the case of axial fans, efficiency can be increased by improving the shroud [36].

## Conclusion

The energetic performances and the flow of an innovative hybrid air extractor operating at low pressures are investigated using steady RANS simulations. It is the first time that the flow of a hybrid extractor with the fan placed out of the flow is analyzed. The simulations are performed on a real geometry with a complex shape. Fine grids are used in the calculations. The extractor has three modes of operation that are analyzed with three distinct configurations. First, the pressure drop mode simulations show the flow escapes the duct easily and generates small pressure drop. Then, the static mode simulations show the flow accelerates between the cone and the duct. For duct flow rates corresponding to a Reynolds number below 40,000, the Venturi effect generates a suction into the duct. Finally, in the fan’s dynamic mode simulations, MRF modeling shows the flow below the fan appears to be swirling. Three areas have been highlighted: in area A, at the center, the flow goes up without rotation. In area B, the flow, still going up, rotates. In area C, the flow still rotates and goes down. The pressure drop mode simulation configuration and the static mode simulation configuration give results in good agreement with the experimental data. The simulation results are quite consistent with the experimental ones. The behavior of the hybrid extractor is well predicted and the simulations give information of the effect of the geometry on the flow.

The energetic evaluation of the hybrid air extractor shows that, almost 80% of the time, the extractor can work without electric power consumption. It confirms the interest of this kind of extractor with a fan placed out of the duct. However, the lack of efficiency in electric mode implies that shape optimization of the fan and the guard are required for this extractor in order to be competitive with classical centrifugal fans.

## Acknowledgements

This work was granted access to the HPC resources of CINES under the allocation 2013-c20132a7123 made by Grand Equipement National de Calcul Intensif (GENCI). This research has been possible thanks to OpenFOAM developers and community.

## References

Khan, N. , Su, Y. , and Riffat, S. B. , 2008, “ A Review on Wind Driven Ventilation Techniques,” Energy Build., 40(8), pp. 1586–1604.
Ismail, M. , and Abdul Rahman, A. , 2010, “ Comparison of Different Hybrid Turbine Ventilator (HTV) Application Strategies to Improve the Indoor Thermal Comfort,” Int. J. Environ. Res., 4(2), pp. 297–308.
Lai, C.-M. , 2003, “ Experiments on the Ventilation Efficiency of Turbine Ventilators Used for Building and Factory Ventilation,” Energy Build., 35(9), pp. 927–932.
Revel, A. , 1998, “ Testing of Two Wind Driven Roof Ventilators,” INSEARCH Limited, Sydney, Australia, Technical Report No. E98/42/041.
Khan, N. , Su, Y. , Riffat, S. B. , and Biggs, C. , 2008, “ Performance Testing and Comparison of Turbine Ventilators,” Renewable Energy, 33(11), pp. 2441–2447.
De Gids, W. , and Den Ouden, H. P. L. , 1987, “ Three Investigations of the Behavior of Ducts for Natural Ventilation in Which an Examination is Made of the Influence of Location and Height of the Outlet, of the Built-Up Nature of the Surroundings and of the Form of the Outlet,” TNO, The Hague, The Netherlands.
Hughes, B. R. , and Abdul Ghani, S. A. A. , 2008, “ Investigation of a Windvent Passive Ventilation Device Against Current Fresh Air Supply Recommendations,” Energy Build., 40(9), pp. 1651–1659.
Hughes, B. R. , and Abdul Ghani, S. A. A. , 2009, “ A Numerical Investigation Into the Effect of Windvent Dampers on Operating Conditions,” Energy Build., 44(2), pp. 237–248.
Hughes, B. R. , and Abdul Ghani, S. A. A. , 2010, “ A Numerical Investigation Into the Effect of Windvent Louvre External Angle on Passive Stack Ventilation Performance,” Energy Build., 45(4), pp. 1025–1036.
Serag-Eldin, M. A. , 2009, “ Prediction of Performance of a Wind-Driven Ventilation Device,” J. Wind Eng. Ind. Aerodyn., 97(11–12), pp. 560–572.
Kim, J.-H. , Kim, J.-W. , and Kim, K.-Y. , 2011, “ Axial-Flow Ventilation Fan Design Through Multi-Objective Optimization to Enhance Aerodynamic Performance,” ASME J. Fluids Eng., 133(10), p. 101101.
Pfeiffer, A. , Dorer, V. , and Weber, A. , 2008, “ Modelling of Cowl Performance in Building Simulation Tools Using Experimental Data and Computational Fluid Dynamics,” Build. Environ., 43(8), pp. 1361–1372.
Van Hooff, T. , Blocken, B. , Aanen, L. , and Bronsema, B. , 2011, “ A Venturi-Shaped Roof for Wind-Induced Natural Ventilation of Buildings: Wind Tunnel and CFD Evaluation of Different Design Configurations,” Build. Environ., 46(9), pp. 1797–1807.
Van Hooff, T. , Blocken, B. , Aanen, L. , and Bronsema, B. , 2012, “ Numerical Analysis of the Performance of a Venturi-Shaped Roof for Natural Ventilation: Influence of Building Width,” J. Wind Eng. Ind. Aerodyn., 104–106, pp. 419–427.
Blocken, B. , Van Hooff, T. , Aanen, L. , and Bronsema, B. , 2011, “ Computational Analysis of the Performance of a Venturi-Shaped Roof for Natural Ventilation: Venturi-Effect Versus Wind-Blocking Effect,” Comput. Fluids, 48(1), pp. 202–213.
Montazeri, H. , Montazeri, F. , Azizian, R. , and Mostafavi, S. , 2010, “ Two-Sided Wind Catcher Performance Evaluation Using Experimental, Numerical and Analytical Modeling,” Renewable Energy, 35(7), pp. 1424–1435.
Montazeri, H. , 2011, “ Experimental and Numerical Study on Natural Ventilation Performance of Various Multi-Opening Wind Catchers,” Build. Environ., 46(2), pp. 370–378.
Lien, S.-T. J. , and Ahmed, N. A. , 2010, “ Numerical Simulation of Rooftop Ventilator Flow,” Build. Environ., 45(8), pp. 1808–1815.
Farahani, A. , Adam, N. , and Ariffin, M. , 2010, “ Simulation of Airflow and Aerodynamic Forces Acting on a Rotating Turbine Ventilator,” Am. J. Eng. Appl. Sci., 3(1), p. 159.
BSI, 2005, “ Ventilation for Buildings. Performance Testing of Components/Products for Residential Ventilation. Cowls and Roof Outlet Terminal Devices,” British Standards Institution, London, Standard No. EN-13141-5.
ISO, 2008, “ Industrial Fans. Performance Testing Using Standardized Airways,” International Organization for Standardization, Geneva, Switzerland, Standard No. ISO-5801.
Ferziger, J. H. , and Perić, M. , 1996, Computational Methods for Fluid Dynamics, Vol. 3, Springer, Berlin.
Shih, T.-H. , Zhu, J. , and Lumley, J. L. , 1993, “ A Realizable Reynolds Stress Algebraic Equation Model,” NASA Lewis Research Center, Cleveland, OH, Technical Report No. NASA-TM-105993.
Strasser, W. , 2009, “ Cyclone-Ejector Coupling and Optimisation,” Prog. Comput. Fluid Dyn., 10(1), pp. 19–31.
Launder, B. E. , and Spalding, D. B. , 1974, “ The Numerical Computation of Turbulent Flows,” Comput. Methods Appl. Mech. Eng., 3(2), pp. 269–289.
Liu, H.-L. , Ren, Y. , Wang, K. , Wu, D.-H. , Ru, W.-M. , and Tan, M.-G. , 2012, “ Research of Inner Flow in a Double Blades Pump Based on Openfoam,” J. Hydrodyn., Ser. B, 24(2), pp. 226–234.
Vanyo, J. P. , 1993, Rotating Fluids in Engineering and Science, Elsevier, Amsterdam, The Netherlands.
Ballesteros-Tajadura, R. , Velarde-Suárez, S. , Hurtado-Cruz, J. P. , and Santolaria-Morros, C. , 2006, “ Numerical Calculation of Pressure Fluctuations in the Volute of a Centrifugal Fan,” ASME J. Fluids Eng., 128(2), pp. 359–369.
Ballesteros-Tajadura, R. , Velarde-Suárez, S. , and Hurtado-Cruz, J. P. , 2008, “ Noise Prediction of a Centrifugal Fan: Numerical Results and Experimental Validation,” ASME J. Fluids Eng., 130(9), p. 091102.
Gonzalez, J. , Fernandez, J. , Blanco, E. , and Santolaria, C. , 2002, “ Numerical Simulation of the Dynamic Effects Due to Impeller-Volute Interaction in a Centrifugal Pump,” ASME J. Fluids Eng., 124(2), pp. 348–355.
Corsini, A. , Delibra, G. , and Sheard, A. G. , 2013, “ A Critical Review of Computational Methods and Their Application in Industrial Fan Design,” ISRN Mech. Eng., 2013, p. 625175.
Hamidreza, T. , Masoud, B. , and Mohammad, T. R. , 2012, “ An Investigation on Turbocharger Turbine Performance Parameters Under Inlet Pulsating Flow,” ASME J. Fluids Eng., 134(8), p. 081102.
Tallgren, J. A. , Sarin, D. A. , and Sheard, A. G. , 2004, “ Utilization of CFD in Development of Centrifugal Fan Aerodynamics,” International Conference on Fans, London, Nov. 9–10, Vol. 4, p. 99.
Pluviose, M. , 2004, “ Similitude des turbomachines hydrauliques,” Techniques De L’ingénieur, Saint-Denis, France, Report No. TIB173DUO.
Wang, S. K. , 2001, Handbook of Air Conditioning and Refrigeration, McGraw-Hill, New York.
Neal, D. , and Foss, J. , 2007, “ The Application of an Aerodynamic Shroud for Axial Ventilation Fans,” ASME J. Fluids Eng., 129(6), pp. 764–772.
View article in PDF format.

## References

Khan, N. , Su, Y. , and Riffat, S. B. , 2008, “ A Review on Wind Driven Ventilation Techniques,” Energy Build., 40(8), pp. 1586–1604.
Ismail, M. , and Abdul Rahman, A. , 2010, “ Comparison of Different Hybrid Turbine Ventilator (HTV) Application Strategies to Improve the Indoor Thermal Comfort,” Int. J. Environ. Res., 4(2), pp. 297–308.
Lai, C.-M. , 2003, “ Experiments on the Ventilation Efficiency of Turbine Ventilators Used for Building and Factory Ventilation,” Energy Build., 35(9), pp. 927–932.
Revel, A. , 1998, “ Testing of Two Wind Driven Roof Ventilators,” INSEARCH Limited, Sydney, Australia, Technical Report No. E98/42/041.
Khan, N. , Su, Y. , Riffat, S. B. , and Biggs, C. , 2008, “ Performance Testing and Comparison of Turbine Ventilators,” Renewable Energy, 33(11), pp. 2441–2447.
De Gids, W. , and Den Ouden, H. P. L. , 1987, “ Three Investigations of the Behavior of Ducts for Natural Ventilation in Which an Examination is Made of the Influence of Location and Height of the Outlet, of the Built-Up Nature of the Surroundings and of the Form of the Outlet,” TNO, The Hague, The Netherlands.
Hughes, B. R. , and Abdul Ghani, S. A. A. , 2008, “ Investigation of a Windvent Passive Ventilation Device Against Current Fresh Air Supply Recommendations,” Energy Build., 40(9), pp. 1651–1659.
Hughes, B. R. , and Abdul Ghani, S. A. A. , 2009, “ A Numerical Investigation Into the Effect of Windvent Dampers on Operating Conditions,” Energy Build., 44(2), pp. 237–248.
Hughes, B. R. , and Abdul Ghani, S. A. A. , 2010, “ A Numerical Investigation Into the Effect of Windvent Louvre External Angle on Passive Stack Ventilation Performance,” Energy Build., 45(4), pp. 1025–1036.
Serag-Eldin, M. A. , 2009, “ Prediction of Performance of a Wind-Driven Ventilation Device,” J. Wind Eng. Ind. Aerodyn., 97(11–12), pp. 560–572.
Kim, J.-H. , Kim, J.-W. , and Kim, K.-Y. , 2011, “ Axial-Flow Ventilation Fan Design Through Multi-Objective Optimization to Enhance Aerodynamic Performance,” ASME J. Fluids Eng., 133(10), p. 101101.
Pfeiffer, A. , Dorer, V. , and Weber, A. , 2008, “ Modelling of Cowl Performance in Building Simulation Tools Using Experimental Data and Computational Fluid Dynamics,” Build. Environ., 43(8), pp. 1361–1372.
Van Hooff, T. , Blocken, B. , Aanen, L. , and Bronsema, B. , 2011, “ A Venturi-Shaped Roof for Wind-Induced Natural Ventilation of Buildings: Wind Tunnel and CFD Evaluation of Different Design Configurations,” Build. Environ., 46(9), pp. 1797–1807.
Van Hooff, T. , Blocken, B. , Aanen, L. , and Bronsema, B. , 2012, “ Numerical Analysis of the Performance of a Venturi-Shaped Roof for Natural Ventilation: Influence of Building Width,” J. Wind Eng. Ind. Aerodyn., 104–106, pp. 419–427.
Blocken, B. , Van Hooff, T. , Aanen, L. , and Bronsema, B. , 2011, “ Computational Analysis of the Performance of a Venturi-Shaped Roof for Natural Ventilation: Venturi-Effect Versus Wind-Blocking Effect,” Comput. Fluids, 48(1), pp. 202–213.
Montazeri, H. , Montazeri, F. , Azizian, R. , and Mostafavi, S. , 2010, “ Two-Sided Wind Catcher Performance Evaluation Using Experimental, Numerical and Analytical Modeling,” Renewable Energy, 35(7), pp. 1424–1435.
Montazeri, H. , 2011, “ Experimental and Numerical Study on Natural Ventilation Performance of Various Multi-Opening Wind Catchers,” Build. Environ., 46(2), pp. 370–378.
Lien, S.-T. J. , and Ahmed, N. A. , 2010, “ Numerical Simulation of Rooftop Ventilator Flow,” Build. Environ., 45(8), pp. 1808–1815.
Farahani, A. , Adam, N. , and Ariffin, M. , 2010, “ Simulation of Airflow and Aerodynamic Forces Acting on a Rotating Turbine Ventilator,” Am. J. Eng. Appl. Sci., 3(1), p. 159.
BSI, 2005, “ Ventilation for Buildings. Performance Testing of Components/Products for Residential Ventilation. Cowls and Roof Outlet Terminal Devices,” British Standards Institution, London, Standard No. EN-13141-5.
ISO, 2008, “ Industrial Fans. Performance Testing Using Standardized Airways,” International Organization for Standardization, Geneva, Switzerland, Standard No. ISO-5801.
Ferziger, J. H. , and Perić, M. , 1996, Computational Methods for Fluid Dynamics, Vol. 3, Springer, Berlin.
Shih, T.-H. , Zhu, J. , and Lumley, J. L. , 1993, “ A Realizable Reynolds Stress Algebraic Equation Model,” NASA Lewis Research Center, Cleveland, OH, Technical Report No. NASA-TM-105993.
Strasser, W. , 2009, “ Cyclone-Ejector Coupling and Optimisation,” Prog. Comput. Fluid Dyn., 10(1), pp. 19–31.
Launder, B. E. , and Spalding, D. B. , 1974, “ The Numerical Computation of Turbulent Flows,” Comput. Methods Appl. Mech. Eng., 3(2), pp. 269–289.
Liu, H.-L. , Ren, Y. , Wang, K. , Wu, D.-H. , Ru, W.-M. , and Tan, M.-G. , 2012, “ Research of Inner Flow in a Double Blades Pump Based on Openfoam,” J. Hydrodyn., Ser. B, 24(2), pp. 226–234.
Vanyo, J. P. , 1993, Rotating Fluids in Engineering and Science, Elsevier, Amsterdam, The Netherlands.
Ballesteros-Tajadura, R. , Velarde-Suárez, S. , Hurtado-Cruz, J. P. , and Santolaria-Morros, C. , 2006, “ Numerical Calculation of Pressure Fluctuations in the Volute of a Centrifugal Fan,” ASME J. Fluids Eng., 128(2), pp. 359–369.
Ballesteros-Tajadura, R. , Velarde-Suárez, S. , and Hurtado-Cruz, J. P. , 2008, “ Noise Prediction of a Centrifugal Fan: Numerical Results and Experimental Validation,” ASME J. Fluids Eng., 130(9), p. 091102.
Gonzalez, J. , Fernandez, J. , Blanco, E. , and Santolaria, C. , 2002, “ Numerical Simulation of the Dynamic Effects Due to Impeller-Volute Interaction in a Centrifugal Pump,” ASME J. Fluids Eng., 124(2), pp. 348–355.
Corsini, A. , Delibra, G. , and Sheard, A. G. , 2013, “ A Critical Review of Computational Methods and Their Application in Industrial Fan Design,” ISRN Mech. Eng., 2013, p. 625175.
Hamidreza, T. , Masoud, B. , and Mohammad, T. R. , 2012, “ An Investigation on Turbocharger Turbine Performance Parameters Under Inlet Pulsating Flow,” ASME J. Fluids Eng., 134(8), p. 081102.
Tallgren, J. A. , Sarin, D. A. , and Sheard, A. G. , 2004, “ Utilization of CFD in Development of Centrifugal Fan Aerodynamics,” International Conference on Fans, London, Nov. 9–10, Vol. 4, p. 99.
Pluviose, M. , 2004, “ Similitude des turbomachines hydrauliques,” Techniques De L’ingénieur, Saint-Denis, France, Report No. TIB173DUO.
Wang, S. K. , 2001, Handbook of Air Conditioning and Refrigeration, McGraw-Hill, New York.
Neal, D. , and Foss, J. , 2007, “ The Application of an Aerodynamic Shroud for Axial Ventilation Fans,” ASME J. Fluids Eng., 129(6), pp. 764–772.

## Figures

Fig. 1

Schematic view of the studied geometry

Fig. 2

Numerical domains for pressure drop mode and dynamic mode (cubic domain with bold bounds), static mode (parallelepipedic domain with thin bounds), and boundary conditions

Fig. 3

Effect of the mesh on the vertical velocity (m s−1) inside the MRF area. The coarse mesh is on the left, the intermediate mesh is at the top, and the fine mesh is on the right. Each cross represents a cell.

Fig. 4

Mesh for pressure drop mode simulations: (a) cut section in the fan and (b) detail of blades

Fig. 5

Dimensionless pressure drop as a function of Reynolds number

Fig. 6

Nondimensionalized velocity field magnitude and nondimensionalized static pressure at Re = 188,462 for pressure drop mode: (a) nondimensionalized velocity (U/Ub) and (b) nondimensionalized pressure ΔPs/(0.5ρUb2)

Fig. 7

Dimensionless static extraction as a function of Reynolds number

Fig. 8

Nondimensionalized velocity field magnitude and nondimensionalized static pressure at Re = 22,772 and U = 8 m s−1 for static mode: (a) U/U and (b) ΔPs/(0.5ρU∞2)

Fig. 9

Dimensionless pressure as a function of Reynolds number

Fig. 10

Efficiency of the extractor as a function of Reynolds number

Fig. 11

Relative velocity (m s−1) for dynamic mode, seen from below

Fig. 12

Streamlines (a), and velocity vectors (b) of the flow below the fan: (a) swirling flow below the fan and (b) flow between the fan and the guard, (z = 1.67 m)

Fig. 13

Line plot of velocities depending of the radius (a), and position of the different part of the extractor following the radius (b): (a) line plot of the vertical velocity and tangential velocity against the radius at z = 1.67 m and (b) schematic view of the extractor

## Tables

Table 1 Mesh study at Re = 62,820 for static mode simulations
Table 2 Grid dependency analysis for dynamic mode at Re ≈ 30,000
Table 3 Mesh parameters
Table 4 Static pressure against Reynolds number and rotation speed for experimental dynamic mode measurements
Table 5 Position of the important geometric characteristics as a function of the radius

## Discussions

Some tools below are only available to our subscribers or users with an online account.

### Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related Proceedings Articles
Related eBook Content
Topic Collections